Encased — AI agent security

Audit deliverables

Exactly what you get from the security audit

Six deliverables, each with a defined format and a clear use. This is what lands with your team at the end of the five days.

01

Findings report

The main written output: every test we ran, what happened, and what to fix first.

Delivered:
Day 5
Format:
PDF, typically 20–40 pages

What's inside

  • One-page executive summary with overall risk rating and the top three issues
  • Tool-and-permission inventory with a blast-radius score per tool
  • Each finding: severity, affected agent, reproduction steps, evidence reference and recommended fix
  • Findings mapped to OWASP Top 10 for LLM Applications categories
  • Prioritised remediation plan split into this-week, this-month and later

How you use it: Share the summary with leadership; hand the findings section to engineering as a work list.

02

Guardrail blueprint policy

Ready-to-apply runtime rules built from what your agents actually did under attack.

Delivered:
Day 4–5
Format:
Encased workspace policy (also exportable as JSON)

What's inside

  • Allow, review and block rules for each risky tool and action pattern found
  • Each rule linked to the finding that justifies it
  • Rules tested against the recorded attack run before handover
  • Notes on which risks need customer-side controls the gateway cannot provide

How you use it: Switch it on in your Encased workspace, or use it as a specification for your own controls.

03

Attack evidence log

The raw record behind every finding, so nothing in the report has to be taken on trust.

Delivered:
Day 5
Format:
Structured export (CSV/JSON) plus workspace history

What's inside

  • Every planted payload and the source it was planted in
  • Each action the agent then attempted, with timestamps
  • The verdict for each action and which control caught it, if any
  • Replayable against new rules to confirm a fix works

How you use it: Attach to internal risk reviews, or replay after changes to prove a weakness is closed.

04

60-minute readout call

A live walkthrough of the results with your engineering and security leads.

Delivered:
Day 5
Format:
Video call, recorded on request

What's inside

  • Walk-through of critical and high findings with live evidence
  • Agreement on fix order and owners
  • Demonstration of the blueprint policy blocking the recorded attacks
  • Open questions from your team

How you use it: Leave the call with a fix plan everyone has agreed to.

05

Live audit workspace

Your own Encased workspace with the blueprint already applied, so protection starts immediately.

Delivered:
From booking
Format:
Encased workspace with 90 days of Growth included

What's inside

  • Blueprint rules loaded and active
  • Review queue for actions held for human approval
  • Activity dashboard and audit trail of every decision
  • Team seats for your engineers and reviewers

How you use it: Keep running your agents behind the rules after the audit ends.

06

30 days of follow-up

Questions about the report or the rules answered for a month after delivery.

Delivered:
After delivery
Format:
Email

What's inside

  • Clarification of any finding or recommendation
  • Advice on adjusting rules as you fix issues
  • Help interpreting the evidence log

How you use it: Get unstuck while your team works through the fixes.

Not included

So there are no surprises about what the $2,500 covers.

  • Rewriting your application code or changing your infrastructure
  • A penetration test of networks, cloud accounts or non-AI systems
  • A SOC 2, ISO 27001 or any other compliance certification — Encased is not independently audited or certified
  • Host-level network blocking, process isolation or credential rotation (customer-side controls)

Get all six for a fixed $2,500

5 days · up to 10 agents · report, policy, evidence, readout and workspace included.